lkml.org 
[lkml]   [2010]   [Nov]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [ANNOUNCE] New utility: 'trace'
From
Date
On Wed, 2010-11-17 at 15:05 +0200, Török Edwin wrote:
> On Wed, 17 Nov 2010 13:51:52 +0100
> Peter Zijlstra <peterz@infradead.org> wrote:
>
> > On Wed, 2010-11-17 at 14:47 +0200, Török Edwin wrote:
> > > I see --call-graph in the trace record -h output, but it doesn't
> > > seem to work on x86_64 for me.
> >
> > If you want to unwind userspace you need to build everything with
> > framepointers -- sadly some people have been lobbying to remove
> > framepointers from all distro builds, even though on x86_64 its nearly
> > free (i386 does have a significant performance benefit, sadly).
> >
> > There's some work on-going to simply copy out the top of stack and let
> > dwarves go wild at it in userspace, but that isn't done yet.
>
> Even without frame-pointers I should see at least the first caller in
> userspace, since that can be obtained by reading the return address of
> the syscall, right?

Right.

> I don't see any difference in the output of 'trace report' thats why
> I'm asking it is implemented at all right now, or I need additional
> flags to see it.

Ah, I haven't actually looked at any of the perf-trace code yet, so I'm
afraid I can't answer that particular question.

> I do see some difference in output of 'perf record -g/perf report'
> though.

ok.


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2010-11-17 14:13    [W:0.080 / U:5.820 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site